Marginal zone lymphoma is associated with dysregulated T-cell immunity that is not altered by ibrutinib-venetoclax treatment

Key Points

• •Relapsed/refractory MZL is associated with widespread upregulation of PD-1 on immune subsets and dysfunctional T and NK cells.
• •Ibrutinib-venetoclax treatment did not reduce elevated T-cell cytokine production, which may affect responses to subsequent immunotherapy.

Visual Abstract

Abstract

Marginal zone lymphoma (MZL) is a rare non-Hodgkin lymphoma. We recently reported that treatment with ibrutinib-venetoclax under the phase 2 AIM study resulted in an overall response rate of 79%. In the current study, we analyzed the immunology of MZL and the effect of ibrutinib-venetoclax treatment. Peripheral blood mononuclear cell samples from 14 patients with relapsed/refractory MZL were collected over 2 years as part of a preplanned analysis. Immune profile was assessed using spectral flow cytometry and functional assays and compared to age-matched healthy donors. Results were correlated with centrally determined positron emission tomography/computed tomography response. Patients with MZL exhibited normal proportions of monocytes, myeloid-derived suppressor cells, and myeloid dendritic cells, with reduced proportions of plasmacytoid dendritic cells (P < .01). Although there was no difference in total proportions of CD4 or CD8 T cells or natural killer (NK) cells at baseline, there was skewing of memory subsets with upregulation of programmed cell death 1 protein (PD-1) on T cells and TIM3 on NK cells. T-cell production of interferon gamma, tumor necrosis factor α, and CD107a was increased sixfold (P < .001), which did not reduce with treatment. Mature NK cell degranulation was elevated at baseline (P < .05) and decreased with treatment. Complete response to treatment was associated with maintenance or expansion of TIM3⁺ mature NK cells (P < .05). To our knowledge, this is the first study to examine peripheral blood immunology in MZL as a single disease group. Given the lack of T-cell repair, immunotherapy protocols using ibrutinib may be less effective in MZL. This trial was registered at www.clinicaltrials.gov as #NCT02471391.

Introduction

Marginal zone lymphoma (MZL) is an indolent B-cell lymphoma that occurs in ∼10% of patients with low-grade lymphoma.¹ Three subtypes of MZL are recognized: splenic, extranodal, and nodal. Although systemic therapy with regimens containing an anti-CD20 monoclonal antibody can induce clinical responses, the disease is incurable in the relapsed and/or refractory (R/R) setting, with limited standard-of-care treatment options.

MZL is characterized by an expansion of clonal B cells lacking expression of CD5 and CD10.² The peripheral blood (PB) immunology of patients with MZL is poorly described, and because of its rarity, it is often grouped with other non-Hodgkin lymphomas in clinical trials.² An analysis of the PB of 7 patients with diagnostic mucosa-associated lymphoid tissue extranodal MZL has shown a reduction in CD4 total and CD4 naive T cells, but not CD8 T cells or natural killer (NK) cells.³ In addition, reduced proportions of regulatory T cells (Treg) have been described.⁴ Furthermore, T-cell receptor analysis has demonstrated that MZL is associated with a highly skewed T-cell repertoire compared to healthy donors and other lymphoma subtypes.^5,6 Within the myeloid lineage, an increased expression of CD14 and CD169 has been described.⁷ A detailed immune profiling of PB subsets in MZL has not yet been examined at disease relapse.

In recent years, targeted therapies for B-cell malignancies including ibrutinib and zanubrutinib, first- and second-generation irreversible inhibitors of Bruton tyrosine kinase (BTK), and venetoclax, a BH3 mimetic inhibitor of B-cell lymphoma 2 (BCL2), provide new avenues for salvage. Although these agents have demonstrated some single-agent activity in MZL,^8,9 in the MZL arm of the prospective AIM study (ClinicalTrials.gov identifier: NCT02471391), 14 patients with R/R MZL received combination ibrutinib-venetoclax, with an overall response rate of 79% and complete response (CR) rate of 29% at 4 months.¹⁰

Both ibrutinib and venetoclax inhibit essential pathways in malignant B cells leading to cell death and have variable effects on PB immune populations. Although venetoclax has been shown to deplete NK cells in mice¹¹ as a result of interleukin-15 (IL-15) stimulation and induced upregulation of BCL2 and MCL1,12, 13, 14 venetoclax had limited effect on T-cell and NK cell proportions when administered to patients with chronic lymphocytic leukemia (CLL) or acute myeloid leukemia.¹⁵ Ibrutinib, via off-target inhibition of IL-2–inducible T-cell kinase (ITK), has been shown to normalize T-cell proliferation, degranulation, and cytokine production in patients with CLL treated long term.16, 17, 18 Furthermore, we have also recently reported on the PB immunology of patients with R/R mantle cell lymphoma (MCL) treated on the AIM study, demonstrating that long-term combination ibrutinib-venetoclax treatment restored T-cell function.¹⁹

In this study, we extended our correlative analysis of the AIM study to the MZL treatment arm and analyzed PB immune subsets over a 2-year period.

Methods

Patient samples

Fourteen patients with R/R MZL were enrolled on the prospective AIM study (ClinicalTrials.gov identifier: NCT02471391) of combination venetoclax and ibrutinib.¹⁰ Patients received ibrutinib 560 mg daily, and after 4 weeks, venetoclax stepwise weekly dose escalation to 400 mg daily was added. Combination therapy was continued until disease progression or unacceptable toxicity. PB was collected at study enrolment (baseline) and at regular intervals to 24 months. Baseline samples from 1 patient with MZL who failed screening were included in the baseline immunology analysis only. Ten age-matched healthy donor PB samples were obtained from the Australian Red Cross Blood Service as approved by the Melbourne Health Human Research Ethics Committee. PB mononuclear cells (PBMC) were isolated using Ficoll-Paque Plus (GE Healthcare, Chicago, IL) and cryopreserved until required. The study was conducted in accordance with the Declaration of Helsinki.

Response classification

Disease reassessment on therapy was as previously described¹⁰ and performed at weeks 4, 16, 28, 40, and 56 using computed tomography (CT) and ¹⁸F-fluorodeoxyglucose–positron emission tomography (PET; weeks 4, 16, and 56) with bone marrow examination and measurable residual disease assessment.

Flow cytometry

Immune profile was assessed using a single-tube 29-antibody panel on a Cytek Aurora spectral flow cytometer, as previously described.^20,21 Briefly, to reduce experimental variation, all samples from a patient were analyzed in a single-flow cytometry experiment. Data were analyzed using FlowJo version 10.10.0 (BD Biosciences). Live cells were gated initially based on the side scatter area and forward scatter area, followed by gating for single cells. Dead cells were excluded from the analysis by positive staining with the fixable viability dye LIVE/DEAD Blue. Lymphocytes were gated from the proportion of live cells. B cells were gated on CD19⁺CD3^- cells. Conventional CD3⁺ T cells were defined from the population of CD19^- cells. Memory phenotypes were determined by assessing the expression of specific markers on CD4⁺ or CD8⁺ T lymphocytes: naive (CCR7⁺CD45RA⁺), central memory (CM; CCR7⁺ CD45RA^-), effector memory (CCR7^-CD45RA^-), and terminal effector memory (CCR7^-CD45RA⁺). CD3⁻CD56⁺CD16⁺ NK cells were identified, whereas CD3⁺ cells were used to characterize γδ T cells. Subsequently, CD56 and CD16 were used to distinguish various NK subpopulations, including CD56mid16⁺ (mature NK), CD56mid16^- (regulatory NK), and CD56HI (immature NK). Treg were defined as CD25⁺CD127^- within the total CD4⁺ T-cell population. The expression of immune checkpoints Programmed cell Death 1 (PD-1), TIM3, LAG3, 4-1BB, HLA-ABC, and HLA-DR was gated as a percentage of CD4⁺, CD8⁺, Treg, and γδ T cell and NK cell subsets. Myeloid immune subsets were initially gated into 2 populations: live CD33⁺CD11b^- cells and CD33⁺CD11b⁺ cells. Subtypes of monocytes, including classical (CD14⁺CD16^-), intermediate (CD14⁺CD16⁺), and nonclassical (CD14^-CD16⁺), were identified based on their respective percentages. Myeloid dendritic cells (CD11c⁺HLA-DR⁺) were identified from the CD14^-CD16^- population, and plasmacytoid dendritic cell (CD123⁺HLA-DR⁺) subtypes were gated on live cells.

T-cell degranulation and cytokine production assay

PBMC were thawed and incubated overnight at 37°C in 5% carbon dioxide in T-cell media (RPMI 1640, 10% fetal calf serum, 1% volume-to-volume ratio HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid], 1mM sodium pyruvate, 1× GlutaMAX, 1× minimum essential medium nonessential amino acids [all Thermo Fisher Scientific], and 50mM 2-mercaptoethanol [Sigma-Aldrich, St. Louis, MO]) supplemented with 20 IU/mL IL-2 (Miltenyi Biotec, Cologne, Germany). PBMC were incubated with CD2/3/28 T-cell activation beads (Miltenyi Biotec, Cologne, Germany) in the presence of GolgiPlug, GolgiStop, and anti-CD107a allophycocyanin (APC) antibody (BD Bioscience) for 4 hours at 37°C in T-cell media. Cells were stained with LIVE/DEAD Blue (Thermo Fisher Scientific) at room temperature for 20 minutes. Cells were stained for surface T-cell markers (CD3 BV510 [BioLegend], CD8 BV480 [BD Bioscience], CD4 BUV805 [BD Bioscience], CD45RA PerCP Cy5.5, and CCR7 PE-Cy7 [Thermo Fisher Scientific]) in the presence of human Fc block (BD Bioscience) for 30 minutes at 4°C. Cells were fixed/permeabilized for 20 minutes at 4°C in Cytofix/Cytoperm fixation solution (BD Biosciences) and washed twice in Cytofix/Cytoperm Perm/Wash buffer (BD Bioscience). Intracellular cytokine expression was detected by staining with tumor necrosis factor α (TNFα) PE-Cy7 (BD Bioscience), IL-2 BV650 (BD Bioscience), and interferon gamma (IFN-γ) BV786 (BD Bioscience) in Perm/Wash buffer at 4°C for 30 minutes, followed by 2 washes in fluorescence-activated cell sorting (FACS) buffer (2% fetal bovine serum (FBS) in phosphate-buffered saline (PBS)) and fixation in 2% paraformaldehyde. Samples were acquired on an Cytek Aurora spectral flow cytometer and analyzed using FlowJo software (BD Bioscience). The percentage and the geometric mean fluorescence intensity (MFI) of cells positive for IFN-γ, IL-2, TNFα, and CD107a were gated on CD3⁺CD19^- T-cell subsets. Triple cytokine-producing IFN-γ⁺IL-2⁺TNFα⁺ cells were calculated as a percentage of T cells.

NK cell degranulation and cytokine production assay

PBMC were thawed and incubated overnight at 37°C in 5% carbon dioxide in T-cell media supplemented with 20 IU/mL IL-2 (Miltenyi Biotec, Cologne, Germany). PBMC were incubated with NK cell activation/expansion kit (Miltenyi Biotec, Cologne, Germany) in the presence of GolgiPlug, GolgiStop, and anti-CD107a APC antibody (BD Bioscience) for 4 hours at 37°C in T-cell media. Cells were stained with LIVE/DEAD Blue (Thermo Fisher Scientific) at room temperature for 20 minutes. Cells were stained for surface markers (CD3 BV510 [BioLegend], CD16 BV570 [BioLegend], CD57 FITC [BioLegend], and CD56 AF700 [BD Biosciences]) in the presence of human Fc block (BD Bioscience) for 30 minutes at 4°C. Cells were fixed/permeabilized for 20 minutes at 4°C in Cytofix/Cytoperm fixation solution (BD Biosciences) and washed twice in Cytofix/Cytoperm Perm/Wash buffer (BD Bioscience). Intracellular cytokine expression was detected by staining with TNFα PE-Cy7 (BD Bioscience), IL-2 BV650 (BD Bioscience), and IFN-γ BV786 (BD Bioscience) in Perm/Wash buffer at 4°C for 30 minutes, followed by 2 washes in FACS buffer and fixation in 2% paraformaldehyde. Samples were acquired on a Cytek Aurora spectral flow cytometer and analyzed using FlowJo software (BD Bioscience). The percentage and the geometric MFI of cells positive for IFN-γ, IL-2, TNFα, and CD107a were gated on CD3^-CD56⁺ NK cell subsets.

Statistical analysis

Data were analyzed using GraphPad Prism version 6.0 (GraphPad Software, San Diego, CA). Data in dot plots are presented as mean ± standard deviation. An unpaired Mann-Whitney U test was used to compare single-cell populations at baseline to healthy donors. Where multiple-cell subsets are analyzed at baseline or subsets are tracked over time, an unpaired t test with Holm-Šídák correction for multiple comparisons was used. Cytokine production and degranulation were analyzed using an ordinary 2-way analysis of variance with Šídák multiple-comparisons test. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001.

Results

Patients with R/R MZL have dysregulated PB immunity

To examine the PB immunology of patients with R/R MZL, a 29-antibody panel was used to analyze PBMC of the patients (median age, 68 years; range, 38-79) against those of age-matched healthy donors (median age, 66.5 years; range, 23-78). The proportion of B cells in patients with MZL varied from 0.03% to 84.3% of live lymphocytes (Figure 1A), and consistent with circulating disease, 7 of 15 patients had B-cell proportions >40%. Although total monocytes were reduced in some patients, this was not statistically significant (Figure 1B), and there was no change in the proportions of classical, intermediate, and nonclassical monocyte subsets (Figure 1C). The proportions of myeloid-derived suppressor cells and myeloid dendritic cells were normal, whereas plasmacytoid dendritic cells were reduced compared to those in healthy donors (P < .01; supplemental Figure 1). Total NK cell proportions were normal (Figure 1D), but there was a 1.9-fold decrease in regulatory NK cells (P < .01), with no change in mature or immature NK cell subsets (Figure 1E). The proportions of TIM3⁺ immature NK cells and PD-1⁺ regulatory NK cells were increased (P < .05), whereas there was no change in PD-1 or TIM3 expression on mature NK cells (Figure 1F).

Figure 1.

Patients with R/R MZL exhibit significant changes in NK cell subsets but not monocytes. Flow cytometric analysis of patients with R/R MZL (n = 15) and age-matched healthy donors (n = 10): (A) B cells, (B) total monocytes, (C) monocyte subsets, (D) total NK cells, (E) NK cell subsets. PD-1 and TIM3 expression on (F) mature NK cells, (G) immature NK cells, and (H) regulatory NK cells.

Focusing on T cells, there was a reduction in total T cells in patients with MZL (P < .001), which was primarily because of a decrease in CD4 T cells (P < .0001) (Figure 2A). PD-1⁺ CD4 T cells were increased twofold (Figure 2B), with normal proportions of PD-1⁺ CD8 T cells (Figure 2C). The expression of TIM3, LAG3, 4-1BB, HLA-ABC, and HLD-DR was unchanged on CD4 and CD8 T cells (Figure 2B-C). CD4 T-cell memory subsets were altered, with increased proportions of CM cells (P < .001), whereas terminal effector memory CD45RA-positive cells were reduced (P < .01; Figure 2D). PD-1 was upregulated across all CD4 memory subsets, 3.7-fold on naive cells (P < .05), 2.5-fold on CM cells (P < .0001), 1.5-fold on effector memory cells (P < .001), and 2.75-fold on terminal effector memory CD45RA-positive cells (P < .01; Figure 2E). In contrast, for CD8 T cells, only the CM subset was increased (P < .01; Figure 2F), and PD-1 was increased fivefold on naive T cells (P < .001) and twofold on CM T cells (P < .0001) (Figure 2G). The proportions of Treg were unchanged compared to healthy donors (Figure 2H), with a 1.5-fold increase in PD-1⁺ Treg (P < .01; Figure 2I). T-cell receptor γδ (TCRγδ) cells were reduced (P < .05; Figure 2J), with a twofold increase in PD-1⁺ TCRγδ cells (P < .001) and a 3.5-fold increase in TIM3⁺ TCRγδ cells (P < .01; Figure 2K).

Figure 2.

Patients with R/R MZL exhibit significant changes to T-cell phenotype. (A) Flow cytometric analysis of total T cells. Immune checkpoint expression on (B) CD4 T cells and (C) CD8 T cells. (D) CD4 and (E) CD8 T cell memory subsets. PD-1 expression on (F) CD4 and (G) CD8 T cell memory subsets. (H) Total Treg. (I) immune checkpoint expression on Treg. (J) Total γδ T cells and (K) immune checkpoint expression on γδ T cells.

Taken together, these results indicate that R/R MZL is associated with dysregulated T-cell and NK cell subsets, with upregulation of PD-1 and TIM3 suggestive of immune exhaustion.

Long-term treatment with venetoclax-ibrutinib does not change T-cell phenotype

We have previously demonstrated that treatment with ibrutinib and venetoclax is able to promote T-cell repair and normalization of T-cell memory phenotype in MCL,¹⁹ and we therefore explored this in the R/R MZL arm of the study. However, when we examined PBMC collected at weeks 4 and 16 and at 1 year and 2 years of treatment despite clearance of circulating B cells by week 16 (supplemental Figure 2), we did not see any change in T-cell memory subsets or PD-1 expression across CD4 or CD8 T cells, and they remained skewed toward an exhausted phenotype (Figure 3A-B; supplemental Figure 2).

Figure 3.

Treatment with ibrutinib-venetoclax does not change T-cell memory phenotype, cytokine, or degranulation responses. Flow cytometry analysis of (A) CD4 and (B) CD8 T-cell phenotype at baseline, week 4, week 16, 12 months, and 24 months of treatment. Intracellular cytokine and degranulation responses after 4 hours of stimulation with CD3/28 beads were assessed in n = 6 patients with MZL and n = 10 age-matched healthy donors. IFN-γ⁺, IL-2⁺, TNFα⁺, and CD107a⁺ proportions (C) and MFI (D) on total T cells. Frequency of IFN-γ⁺/IL-2⁺/TNFα⁺ T cells at baseline (E) and on treatment (F).

To examine the effect of ibrutinib-venetoclax treatment on T-cell function, degranulation and intracellular cytokine production were measured by flow cytometry after 4 hours of stimulation with CD2/3/28 T-cell activation beads (n = 6). At baseline, the proportion of T cells producing IFN-γ was increased 6.6-fold (P < .001), with TNFα increased 2.3-fold (P < .0001), and degranulation (measured by cell surface expression of CD107a) increased 2.2-fold (P < .0001) (Figure 3C), compared to those in healthy donors. Although there was a trend of increased IL-2⁺ T cells, this was not statistically significant (Figure 3C). In addition, the MFI of TNFα and CD107a in T cells was significantly increased (P < .0001) (Figure 3D). Increased TNFα⁺ and CD107a⁺ populations were maintained across CD4 and CD8 T-cell subsets, with increased IFN-γ⁺ cells seen in CD8 T cells only (supplemental Figure 3). In contrast to our prior findings in MCL, treatment on study did not change either the proportion-positive or MFI of T cells expressing IFN-γ, IL-2, TNFα, or CD107a (supplemental Figure 3). Importantly, the proportion of T cells that coexpressed IFN-γ/IL-2/TNFα was increased 12-fold compared to that in healthy donors (P < .001; Figure 3E), which was unchanged with treatment (Figure 3F). Overall, this indicates that T cells from R/R MZL have dysregulated cytokine and degranulation responses that are unchanged by ibrutinib-venetoclax treatment.

Treatment response is associated with changes to innate immunity

The clinical trial primary end point was response at 16 weeks, which demonstrated a CR rate of 43% by PET-CT.¹⁰ However, immunological changes as a result of ibrutinib treatment take 6 to 12 months,16, 17, 18 and therefore, to identify immunological biomarkers of response, patients were stratified according to PET-CT response at 1 year of therapy into those that achieved CR (n = 6) vs those that did not achieve CR (partial response, n = 4; stable disease, n = 1; or progressive disease, n = 3) and correlated with the immune profile at baseline and at week 4, week 16, 1 year, and 2 years of treatment.

The presence of circulating B cells at baseline did not correlate with treatment response (Figure 4A), with similar rates of circulating disease at baseline and during treatment regardless of clinical response. Consistent with the lack of change in T-cell memory subsets with treatment, there was no correlation between T-cell subsets and disease response (Figure 4B-C; supplemental Figure 4). Although the proportion of PD-1⁺ CD4 and CD8 T-cell memory subsets was significantly increased at baseline in patients with MZL compared to healthy donors, there was no correlation to disease response (Figure 4D-E).

Figure 4.

Neither the presence of circulating tumor cells nor baseline T-cell phenotype correlates with treatment response. Patients were stratified into CR (n = 6) and non-CR (n = 8) based on treatment response at 12 months. Correlations with (A) total B cells, (B) CD4, and (C) CD8 T-cell memory subsets and (D) PD-1⁺ CD4 T cell and (E) PD-1⁺ CD8 T-cell memory subsets.

Focusing on NK cells, there was no change in NK cell subsets with treatment (supplemental Figure 5) or any difference in the proportion of NK cells or NK subsets on the basis of disease response at baseline or during treatment (Figure 5A). TIM3 expression on NK cell subsets at baseline correlated with disease response, with an elevation of TIM3⁺ mature NK cells in those patients not attaining CR (P < .05) compared to healthy donors (Figure 5B). During the first 4 weeks of single-agent ibrutinib, patients who attained eventual CR either maintained or increased their levels of TIM3⁺ mature NK cells, whereas patients who did not attain CR had a fall in this subset (P < .05; Figure 5C). At baseline, there was a significant increase in patients with TIM3⁺ immature NK cells attaining CR compared to healthy donors (P < .01; Figure 5B). However, no significant difference was found between those patients attaining CR compared to those without CR (Figure 5D). Further assessment of NK cell function was undertaken with analysis of degranulation and intracellular cytokine production after CD2/CD335 (NKp46) stimulation. NK cell function in patients (n = 6) was comparable to that in age-matched healthy donors (n = 5), with equivalent proportions of IFN-γ⁺, IL-2⁺, TNFα⁺, and CD107a⁺ NK cells (Figure 5E). Mature CD107a⁺ NK cells were increased at baseline compared to those in healthy donors (P < .05), which decreased with treatment lower than those in healthy donors by week 16 (P < .001; Figure 5F). Although baseline CD107a was higher in TIM3^- NK cells, similar decreases in CD107a with treatment were seen across TIM3⁺ and TIM3^- mature NK cells (supplemental Figure 5). However, there was no correlation between CD107a⁺ total or mature NK cells and disease response (Figure 5G-H). Overall, these data suggest that the NK cell phenotype is altered after ibrutinib-venetoclax treatment, but this does not appear to correlate with MZL response.

Figure 5.

Change in TIM3⁺ mature NK cells, but not NK cell function, upon ibrutinib treatment correlates with long-term treatment response. Patients were stratified into CR and non-CR based on treatment response at 12 months. Correlation with (A) total NK cells, (B) baseline NK cell subsets, (C) fold change in TIM3⁺ mature NK cells from baseline, and (D) fold change in TIM3⁺ immature NK cells from baseline. Intracellular cytokine production and degranulation responses after 4 hours of stimulation with CD2/CD335 beads were assessed in n = 6 patients with MZL and n = 5 age-matched healthy donors. Baseline IFN-γ⁺, IL2⁺, TNFα⁺, and CD107a⁺ proportions in (E) total NK cells and (F) mature NK cells. Correlation between (G) CD107a⁺ NK cells and (H) CD107a⁺ mature NK cells over time stratified on treatment response.

Discussion

In this study, we report on a planned substudy of long-term immunological profiling in patients with MZL treated with ibrutinib-venetoclax on the AIM trial over a 2-year period. This provided the opportunity to explore MZL immunology as a distinct cohort rather than combined with other lymphoma subtypes, revealing the impact of combination BTK/BCL2 inhibition and providing important insights into R/R MZL immunology.

Our findings show that PB immunology in patients with R/R MZL is characterized by significant T-cell skewing toward CM T cells and widespread upregulation of PD-1 across T and NK cells. Although the treatment was effective at clearing circulating tumor cells, in contrast to our prior analysis of patients with R/R MCL treated with ibrutinib-venetoclax, where we saw normalization of T-cell memory subsets and function,¹⁹ T-cell subsets remained skewed and had high levels of cytokine production and degranulation that did not change with treatment. As MZL is often associated with chronic antigen stimulation from bacterial and viral infections,² these data suggest that there may be an ongoing antigen source that is not resolved with tumor eradication.

TIM3 expression was elevated on immature NK cells, especially in patients who achieved CR. Upon ibrutinib treatment, there was no change in TIM3⁺ immature NK cells, but TIM3⁺ mature NK cells increased in CR patients, suggesting increased NK cell maturation, which did not occur in patients that did not achieve CR. TIM3 has previously been described to activate or suppress NK cell cytotoxicity depending on the context.²² In addition, we have previously described that ibrutinib inhibits NK cell degranulation because of off-target inhibition of ITK in CLL.²³ The higher rate of baseline degranulation in TIM3^- mature NK cells and the equal impact of ibrutinib-venetoclax treatment in TIM3⁺ and TIM3^- populations suggest that, in this instance, TIM3 acted as an inhibitor of mature NK cell degranulation, which was further suppressed by ibrutinib. However, the analysis is limited by the number of patients, especially once stratified by response, and these results require validation in larger cohorts.

To our knowledge, this is the first study to describe R/R MZL PB immunology as a distinct cohort, rather than being combined with the analysis of other non-Hodgkin lymphomas. Given that we have demonstrated that MZL and MCL had different immunological responses to ibrutinib-venetoclax, future studies exploring how PB disease immunology within distinct lymphoma subtypes affects treatment response are warranted.

As immunotherapy with bispecific antibodies or chimeric antigen receptor T (CAR-T) cells is increasingly being applied in the care of patients with R/R B-cell lymphoma, including MZL,24, 25, 26 the aberrant and noncorrected immunology in patients with MZL may indicate a lower probability of durable responses in this lymphoma subtype. Furthermore, the elevated T-cell cytokine responses demonstrated in our data are likely contributors to cytokine release syndrome and/or immune effector cell-associated neurotoxicity syndrome in this patient population, including the high rate of immune effector cell-associated neurotoxicity syndrome observed in patients with MZL treated with axicabtagene ciloleucel CAR-T on the ZUMA-5 clinical trial.²⁷ Further, it suggests that combination immunotherapy protocols utilizing BTK inhibitors before bispecific antibodies or CAR-T may not provide the intended immunomodulation in MZL that has been found to be beneficial in other lymphoma cohorts,^21,28,29 and alternate immunomodulatory drug strategies should be explored to increase efficacy and mitigate potential immunotoxicity.

Conflict-of-interest disclosure: S.M.H. has received honoraria from BeiGene, AstraZeneca, and Janssen; and has served on advisory boards for BeiGene and Eli Lilly. C.S.T. has received honoraria from BeOne, Janssen, AstraZeneca, Merck, and AbbVie; and research funding from BeOne, AbbVie, and Janssen. M.A.A. received honoraria from AbbVie, Janssen, AstraZeneca, BeiGene, CSL Behring, Takeda, Sobi, Specialised Therapeutics, Novartis, Roche, and Gilead Sciences; and is an employee of the Walter and Eliza Hall Institute of Medical Research, which receives milestone payments related to venetoclax, and she is entitled to a share in that payment. D.S.R. has received research grants from Bristol Myers Squibb, Novartis, and Amgen; and has served as an advisory board member for Specialised Therapeutics, Takeda, CSL Behring, Jazz Pharmaceuticals, and Cellenkos. The remaining authors declare no competing financial interests.

The current affiliation for L.H.H.L. is Murdoch Children’s Research Institute and the Novo Nordisk Foundation Center for Stem Cell Medicine, reNEW, Melbourne, Australia.